Status: Finalised Internal Document id: DBAD a Page 1 (25) By Christian Hultholm,

Transcription

1 Status: Finalised Internal Document id: DBAD a Page 1 (25) POWER-GEN Asia 2015 Optimising a generation fleet with a heavy penetration of renewables utilising the Dispatch Simulator By Christian Hultholm, Content Manager, Marketing & Business Development, Power Plants From Wärtsilä

2 Status: Finalised Internal Document id: DBAD a Page 2 (25) Table of Contents Abbreviations and definitions... 3 Abstract... 5 Chapter 1 Introduction power systems in change... 6 Chapter 2 Background for the Dispatch Simulator... 7 Chapter 3 Operating a power system The basics of a power system Dispatching of power... 8 Chapter 4 Key messages of the Dispatch Simulator Chapter 5 The power system in the Dispatch Simulator Basic setup Power generation fleet performance Operation philosophy Chapter 6 Operation of the Dispatch Simulator Demand vs. supply Decisions and control Monitoring and information Chapter 7 Outcome of the Dispatch Simulator Chapter 8 The impact of Smart Power Generation Bibliography... 22

3 Status: Finalised Internal Document id: DBAD a Page 3 (25) Abbreviations and definitions Abbreviations CCGT EIA ICE IEA REN21 SPG Combined cycle gas turbine U.S. Energy Information Administration Internal combustion engine International Energy Agency Renewable Energy Policy Network for the 21st Century Smart Power Generation Definitions Blackout An unintentional and complete interruption of the electricity supply in a given service area, caused by a severe mismatch in electricity supply and demand. Capacity factor The ratio between the actually produces electric energy and the theoretical, maximum amount of electricity which could have been produces during a certain time span (used for non dispatchable, i.e. Intermittently producing power plants) Dispatch Simulator Load factor Load shedding Merit order Minimum down-time A tool for modelling a large scale power system and giving the user the opportunity to try different ways of achieving an optimum dispatching of power. Its main purpose is to serve as a means of demonstrating and educating about power system challenges and the related solutions. The ratio between the current output and the maximum output of a power plant. A partial interruption of the electricity supply in a given area, caused by a mismatch in electricity supply and demand. Usually a deliberate action taken by the power system operator in order to prevent the situation from escalating into a full scale blackout. The principle of ranking available power generation assets based on their marginal costs of production, so that those with the lowest costs are the first ones to be brought online to meet demand (base load), and the plants with the highest costs are the last to be brought online (peak load). The shortest time a power plant has to remain shut down, after shutting it down. Only after the minimum down-time has passed it can be started again.

4 Status: Finalised Internal Document id: DBAD a Page 4 (25) Minimum uptime Net load Non-spinning service Spinning reserve Short-run marginal cost Smart Power Generation The shortest time a power plant has to be run after starting it. Only after the minimum up-time has passed, it can be shut down. The total electricity demand minus renewable generation. This remaining part of the demand has to be met with power generation that can be dispatched, i.e. generating units that can be ramped and/or started and sopped as needed. Off-line capacity available to come on line in the event of a contingency; must satisfy requirements for start-up time and ramping capability. Online contingency reserve; synchronised to the grid. The price required covering the fuel costs, as well as operating and maintenance costs of a power plant. A new technology which enables an existing power system to operate at its maximum efficiency by most effectively absorbing current and future system load variations, hence providing dramatic savings. The technology is based on multiple interconnected internal combustion engines. Legal disclaimer This document is provided for informational purposes only and may not be incorporated into any agreement. The information and conclusions in this document are based upon calculations (including software built-in assumptions), observations, assumptions, publicly available competitor information, and other information obtained by Wärtsilä or provided to Wärtsilä by its customers, prospective customers or other third parties (the information ) and is not intended to substitute independent evaluation. No representation or warranty of any kind is made in respect of any such information. Wärtsilä expressly disclaims any responsibility for, and does not guarantee, the correctness or the completeness of the information. The calculations and assumptions included in the information do not necessarily take into account all the factors that could be relevant. Nothing in this document shall be construed as a guarantee or warranty of the performance of any Wärtsilä equipment or installation or the savings or other benefits that could be achieved by using Wärtsilä technology, equipment or installations instead of any or other technology.

5 Status: Finalised Internal Document id: DBAD a Page 5 (25) Abstract The Dispatch Simulator is an application that shows the complexity of large-scale power systems. It demonstrates how substantial amounts of intermittent renewable power affect the system and the challenges that arise. Furthermore, it gives the user the unique possibility to try different strategies and generation technologies for managing the real-time balancing of electricity supply and demand. With the Dispatch Simulator anyone can act as a power system operator and take decisions on how to operate the different generation assets in a large-scale grid. The Dispatch Simulator illustrates in a concrete and understandable way the challenges of keeping the grid in balance. Furthermore, both the total generating costs and CO2 emissions are key parameters which the user continuously monitors and try to minimise. The Dispatch Simulator also gives the user the opportunity to add certain amounts of wind and solar power to the generation fleet, as well flexible gas-fired capacity. By utilising this asset for the balancing, the Dispatch Simulator demonstrates how the use of the larger power plants in the fleet can optimised, as these can be operated in a more stable manner. This paper will review the challenges modern power systems face and different approaches for tackling these issues. With the help of the Dispatch Simulator such future scenarios can be easily and rigorously presented and analysed.

6 Status: Finalised Internal Document id: DBAD a Page 6 (25) Chapter 1 Introduction power systems in change Globally, energy consumption is forecasted to continue growing at a steady pace during the next two decades, according to the report World Energy Outlook 2012 by the International Energy Agency (2012), henceforth IEA. The main driver for the increasing demand is the development of the emerging economies, such as China, India and the Middle East countries. With approximately 1.3 billion people still lacking electricity, there is no reason to anticipate any short or mid-term changes in this trend. (IEA 2012: 1 7) At the same time, concerns related to increased emissions, global warming and depleting fossil fuel resources are receiving ever-increasing focus. This has led to political pressure and efforts to reduce the dependency on these traditional power sources. As a direct result, the growth of emission-free, renewable power has been tremendous during the last decade. Interestingly, the strong demand for a higher share of renewable energy has not even been set back by the financial crisis, but has on the contrary continued to increase, as stated in a report from Renewable Energy Policy Network for the 21st Century (2013), hereafter REN21. The year of 2012 signified a substantial milestone in the history of renewables, as they accounted for more than half of the increase in electric generating capacity. This trend is expected to continue and even accelerate further. (IEA 2012: 6; REN : 19 23) With an increasing share of intermittent, renewable power generation, new challenges arise. From an economical point of view, these issues can even be in the trillion euro range (How to 2013). The integration of the renewables into existing power systems, leads to increasingly challenging load profiles for the traditional thermal power plants. In order to maintain system stability, the fossil-fired plants must balance any unforeseen deviations in output from the renewables at all times. In practice, this means more starts and stops, increased cycling and more operation on partload. The dominating thermal plants, based on steam and gas turbines, are not well suited for such operation. As a result, both fuel consumption and maintenance costs will increase considerably. (Arima 2012: 6 8; Engblom 2014: 16 23; IEA 2014a: ; Klimstra 2014: 63 67; Rautkivi 2014: 14 15; van den Bergh, Delarue & D haeseleer 2013: 1 8) The increased need for different forms of flexibility in power systems is widely acknowledged in the energy business today, and flexible power plants are considered one of the cornerstones for providing this flexibility (Appleyard 2014: 18; IEA 2014a: ; IEA 2014b: ). Consequently, power plants which are able to provide the necessary, flexible balancing power, without suffering from the heavily varying operation profile, will be essential in modern power systems. This paper will review the challenges modern power systems face and different approaches for tackling these issues. With the help of the Dispatch Simulator various future scenarios can be easily and rigorously presented and analysed.

7 Status: Finalised Internal Document id: DBAD a Page 7 (25) Chapter 2 Background for the Dispatch Simulator Tell me and I will forget; show me and I may remember; involve me and I will understand. - Chinese proverb This old piece of Chinese wisdom summarises the purpose of the Dispatch Simulator. Power systems are the largest and most complex man-made dynamic systems that exist (Hultholm 2013a: 40; V. Riihimäki, meeting, 28 Oct 2013). Therefore, explaining different concepts and models for optimising power systems to laymen is a difficult task. That is where the interactive tool enters the picture. Power systems are considered one of the most essential infrastructures on a national level. Behind their importance lies a vast array of economic, social and technical variables. From an economic point of view, most industries rely heavily on a reliable and pricewise competitive power supply. The social aspect is important, since power has become an essential part of life for most people in the world. Finally, technically power systems are unique in the sense that they require continuous balancing, as electricity cannot be efficiently stored in large quantities. (Ray 2007: ; Rebennack, Pardalos, Pereira & Iliadis 2010: vii) Due to the great significance of power systems for society, the Dispatch Simulator will provide the means for explaining the key challenges in a manner understandable to laymen, as well as for presenting different solutions to those issues. Briefly, this tool shall illustrate a large-scale power system and the various forms of power plants commonly used: coal-fired plants, gas turbine plants, hydro plants and nuclear plants. In addition to this supply, a typical load profile will complete the basic setup. The user will function as the dispatcher, i.e. manage the whole system both in terms of stability and feasibility. The simulator will then show how the whole system is affected by the introduction of increasing amounts of intermittent renewable power in the form wind and solar power. Finally, the impact and value of adding Smart Power Generation (SPG) to the generation mix will be highlighted. SPG is a new technology which enables an existing power system to operate at its maximum efficiency by most effectively absorbing current and future system load variations, hence providing dramatic savings. The technology is based on multiple interconnected internal combustion engines (ICE). Next, the basics of a power system and its operation will be described. After that, a systematic review of the Dispatch Simulator and its possibilities will follow. Chapter 3 Operating a power system This chapter will briefly introduce the key aspects of how a large-scale power system is set up and how it is operated. Finally, the simplified model of a power system developed for the Dispatch Simulator is described.

8 Status: Finalised Internal Document id: DBAD a Page 8 (25) 3.1 The basics of a power system Briefly, an electric power system can be described as real-time energy delivery system. The electricity demand (aka load) and the supply (aka generation) are constantly matched, meaning that generators produce the power as demand calls for it, typically without storage possibilities. The main sections of a power system are generation, transmission and distribution. (Blume 2007: 3 139; Eirgrid 2009: 1; Ray 2007: ) If the demand at some point exceeds the supply, the frequency of the power system will start to decrease (Blackburn & Domin 2014: 269; Blume 2007: 169). A frequency change immediately affects sensitive equipment, such as electric motors used in the industry, so therefore keeping a stable frequency is the main goal in a power system (Klimstra & Hotakainen 2011: 78). Oppositely, if the demand is lower than the supply, the frequency will begin to increase (Klimstra 2014: 22). Such over-frequency conditions are usually caused by a sudden loss of load (Blackburn & Domin 2014: 269). In case of an abnormally decreasing frequency, under-frequency relays will automatically shed load in order to prevent the frequency from reaching critical levels. Simply put, load shedding means that the supply to some of the consumers is temporarily stopped. This way it is possible to quickly match supply and demand again, hence mitigating an escalating situation. (Blume 2007: ) However, extreme load variations may cause the power system to become unstable, which may results in a full-scale blackout. Such an outage is unintentional and means that a complete interruption of the electricity supply has taken place. Recovering from such a calamity is a major and complicated undertaking (Klimstra & Hotakainen 2011: 149). (Blume 2007: ) The Dispatch Simulator will focus exclusively on the generation part of the power system. The most common power plant technologies on a global scale will be part of the simulator: coal-fired power, gas-fired power, hydro power, nuclear power, wind power and solar power (GlobalData 2014). 3.2 Dispatching of power In order to continuously maintain the delicate balance between power demand and supply, the power system operator is monitoring both of these parameters and striving to maintain equilibrium between them. In order to master this process, commonly known as dispatching, the system operator (aka the dispatcher) is involved in three key undertakings: forecasting, planning and controlling (Arima 2012: 4). Forecasting involves estimations of the future load and the available generation assets. This is done for different time horizons, and naturally the longer ahead the forecasting is done, the greater is the risk for deviations. However, fairly precise results can be attained with the help of modern software in combination with experience. (Arima 2012: 4; Blume 2007: ; Singh 2008: 51) An especially important time horizon is the four hours ahead forecast, since that in most cases is the maximum time it take to synchronise a large combined cycle gas turbine plant (CCGT) (Strbac,

9 Status: Finalised Internal Document id: DBAD a Page 9 (25) Aunedi, Pudjianto, Djapic, Teng, Sturt, Jackravut, Sansom, Yufit & Brandon 2012: 22). Such plants are commonly used for providing the power system reserves (DNV KEMA 2013: 3; Rautkivi 2014: 14; Redpoint 2013: 6 7). As further explained in chapter 4, one of the key messages of the Dispatch Simulator, is to convey how SPG enables savings through fast-starting, nonspinning reserves. The load forecasting accuracy is typically in the range of 1 3% for the next day (Feinberg & Genethliou 2005: 270). Naturally, the average forecasting error for four hours ahead will be smaller, but still an important factor to consider. Moreover, part of the forecasting process are also estimations of the wind and sun, and consequently of the output from these intermittent renewable power sources. Even though the forecasting tools have improved a lot, recent studies by e.g. Imperial College London have shown that the four hour ahead forecasting error of wind still is approximately 15% (Strbac et al. 2012: 22). Based on the forecasting, the system operator does the actual planning of how the generation assets in the power system will be used. The typical process for ranking the dispatch order of the power plants is the so called merit order. By first arranging the assets according to their short-run marginal costs 1 of production, those with the lowest costs are then the first ones to be brought online to meet the demand, so called baseload plants. Consecutively, plants with increasingly higher costs are brought online, ultimately ending with peaking plants. (Arima 2012: 4; Blume 2007: ) Naturally, renewables are the first in the merit order, since they do not have any fuel costs and hence their short-run costs are close to zero. Typically, the renewables are followed by nuclear and coal plants which normally have low marginal costs. Next, large CCGTs are brought online, followed by smaller open cycle gas turbines and internal combustion engines. (Arima 2012: 4; IEA 2014b: 29 30) However, large thermal units, such as nuclear, coal-fired and CCGT power plants, have long startup times. This means that the so called unit commitment, i.e. the decision whether a plant will online during a certain period, must be made well in advance. Depending on the power plant technology, it may even be done in the day-ahead timeframe. (Hodge, Florita, Orwig, Lew & Milligan 2012: 2) In addition to the start-up times of the power plants, the crucial parameters which the dispatcher has to consider also include e.g. minimum up-times and minimum down-times. The former is the shortest time a power plant has to be run after starting it. Only after the minimum up-time has passed, it is possible to shut down the plant again. The minimum down-time depicts the shortest time a power plant has to remain shut down, after it has been fully unloaded. Only after the minimum down-time has passed, it can be started again. (Arima 2012: 4) 1 The price required to cover the fuel costs, as well as operating and maintenance costs of a power plant.

10 Status: Finalised Internal Document id: DBAD a Page 10 (25) Finally, the dispatcher is also responsible for maintaining adequate power system reserves. These are then activated in case of forecast errors or calamities, as further described in chapter 4. (Arima 2012: 4) In addition to the forecasting and planning activities, the dispatcher is also continuously involved in controlling duties. The focus of these activities is to keep the frequency stable in the whole grid, as well as the voltage steady locally. The routine tasks normally revolve around the dispatch plan, including the load forecasts and merit order. (Arima 2012: 5) To sum up, a dispatcher s work can briefly be described as reliably delivering energy to the customers at the best possible price (Arima 2012: 5; Eirgrid 2009: 1 2: Eskola 2008: 27). Hence, both of these two aspects shall be cornerstones of the Dispatch Simulator and the frame for its delivery of the key messages. Chapter 4 Key messages of the Dispatch Simulator Based on the studies conducted by DNV KEMA (2013) and Redpoint (2013), as well as on the research of Dr. Klimstra (Klimstra & Hotakainen 2011; Klimstra 2014), a number of key messages in terms of quantifying the value of Smart Power Generation can be defined. These messages are especially highlighted in the Dispatch Simulator, through its features. The messages are that gasfired ICE power plants can contribute to optimising power systems thanks to efficient part-load operation, fast start-up and ramp-up times, the ability to provide non-spinning power system reserves, the capability to cope with frequent cycling negligible cost impact, and the potential to improve the efficiency of the other power plants in the generation fleet. Efficient part-load operation First of all, power plants based on ICE consist of multiple relatively small units. Having several smaller units instead of a few large ones gives several advantages. All thermal machines suffer in terms of efficiency at low loads, but the low loads can be avoided through the utilisation of many smaller units. From a dispatching point of view, this means that by switching units on and off, the total output of the plant can cleverly be adapted to meet the actual load. By keeping the individual units highly loaded, i.e. close to their nominal output, the efficiency can be kept high throughout the whole load range. (DNV KEMA 2013: 12; Klimstra & Hotakainen 2011: ; Redpoint 2013: 9) Fast start-up and ramp-up Another important feature of ICEs is their fast starting and ramping capability. Typically, such power plants can be started and synchronised to the grid within half a minute and their full output can be reached within five minutes from the start-up command. Looking at the start-up and rampup times of gas turbines, they are normally multiples of the previously mentioned times. Other thermal power plants, such as coal-fired ones, often require several hours of preparations. (DNV KEMA 2013: 12; Klimstra & Hotakainen 2011: ; Redpoint 2013: 6)

11 Status: Finalised Internal Document id: DBAD a Page 11 (25) From a dispatching point of view, the benefit of agile power plants is that they can be started just minutes before they are actually need and still be brought online timely. Less flexible power plants need, on the other hand, to be started well ahead, in order to be available when they are needed. That of course comes at a cost, since they are basically run before the need is there, and most likely then also at a low load (since all the energy they produce might not yet be needed), and thereby at a low efficiency. (Franck & Hägglund 2014: 62) Non-spinning power system reserves All utilities are required to keep a certain amount of reserve generation capacity available. In case of a calamity, such as a power plant trip or a transmission line failure, these reserves are started. Other situations, in which the reserves are called upon include forecast errors, either of the load or the renewable output (as described more in detail in chapter 3.2). (Blume 2007: 184; Shaalan 2000: 8.19) Typically, a substantial share of these reserves must be kept spinning, i.e. synchronised to the grid, operating on part-load, in order to be able meet the required response times (Blume 2007: ; Eirgrid 2009: 5; Klimstra 2014: 49). As described earlier, part-load operation has a negative impact on the efficiency. However, the fast start-up and ramp-up times of gas-fired ICE power plants brings an especially interesting benefit from a power system reserve point of view. Technically, it is possible to keep such plants offline, in standby-mode, and still be able to meet the response times of a considerable share of the reserves. The advantage of such a setup is that there is no fuel consumption while in stand-by, nor are any emissions generated. Moreover, since the plants are not running, they do not suffer from wear either. (Hultholm 2014: 37; Klimstra 2014: 50) Capability to cope with frequent cycling with negligible cost impact Another aspect of flexibility which is characteristic for ICE power plants is the fact that they are able to cope with frequent cycling, i.e. starting, ramping and stopping, without incurring any additional maintenance needs. Unlike competing technologies such as gas turbines, not even rapid changes in operation increase the variable operation and maintenance costs. (DNV KEMA 2013: 1; Klimstra & Hotakainen 2011: 143; Klimstra 2014: ; Redpoint 2013: 9) Moreover, when considering the start-up costs of a power plant, the fuel burnt in the start-up process normally accounts for a significant part. However, in comparison to e.g. gas turbines, ICE power plants have a very modest consumption of start-up fuel, as a result of the fast start-up process. (Kujala & Purohit 2013: 4 10) Improving the efficiency of other power plants in the generation fleet The maybe single most important message which is to be communicated through the Dispatch Simulator is how gas-fired ICE power plants can contribute to improving the overall power system efficiency. As these plants can provide a considerable part of the reserves (non-spinning), as well as quickly do short-term balancing during peak hours, the operation of the other thermal power plants in the generation mix can be kept more steadily. Thus, as those plants are relieved of such duties, they can be utilised to a higher extent for pure energy production, i.e. run at a higher load

12 Status: Finalised Internal Document id: DBAD a Page 12 (25) and thereby at a higher efficiency. Moreover, since the need for cycling the other thermal power plants decreases, also the start-up related costs of the plants are reduced. (DNV KEMA 2013: 24; Hultholm 2014: 39 41; Kujala & Purohit 2013: 19 20; Redpoint 2013: 11 12) To sum up, the core messages which are to be communicated through the Dispatch Simulator are that gas-fired ICE power plants (SPG) can contribute to optimising power systems on the account of the following abilities: Efficient part-load operation - substantial amounts of intermittent renewable power can be efficiently integrated in an existing power system. Fast start-up and ramp-up -sudden variations in output for renewables can be cleverly managed. Provision of non-spinning power system reserves - the amount of costly spinning reserves can be substantially reduced when introducing fast-starting stand-by reserves. Capability to cope with frequent cycling with negligible cost impact - the cycling of coal and gas turbine power plants and their related maintenance costs can be reduced. Improving the efficiency of other power plants in the generation fleet - coal and gas turbine power plants can be utilised more efficiently by running them at a higher load factor. Chapter 5 The power system in the Dispatch Simulator The process of deciding which power plants to dispatch is in practice extremely complex, with a vast number of variables which must be taken into account (Blume 2007: 195; Eirgrid 2009: 1; Eskola 2008: 24). However, in the Dispatch Simulator it is crucial to simplify this aspect, in order to keep the focus on the key messages. 5.1 Basic setup The three basic components of the power system in the Dispatch Simulator are: The power generation fleet Wind and sun profiles The load profile The power generation fleet in the simulator initially has a total capacity of 11 GW, corresponding to the national capacity of e.g. Singapore (EIA 2011). This capacity is distributed among the different, major power plant technologies according to a worldwide average. All the most commonly used technologies are represented: coal-fired plants, gas-fired plants, hydro plants and nuclear plants. In order to simplify the setup, the coal-fired capacity is assumed to use a combination black coal and lignite. Furthermore, the whole gas-fired capacity is represented by

13 Status: Finalised Internal Document id: DBAD a Page 13 (25) CCGTs, which have been the primary technology of choice for gas power plants since the early 1990s (IEA 2010: 1). The performance of the different plants in the simulator is representing average values for existing technology. The performance values are based on a comprehensive review of literature and available equipment specifications. This initial, default setup of the power generation fleet corresponds to a traditional generation mix like the situation used to be some years ago. Such a system is fairly straight forward to operate: the largest plants nuclear, coal-fired ones and hydro provide a steady baseload around the clock. The peak hours, typically in the morning and early evening, are then catered for by gasfired plants. (How to 2013; Klimstra & Hotakainen 2011: 171) Additionally, the user of the Dispatch Simulator has the possibility to add up to three more power plant technologies to the generation mix: wind power (in steps up to 3 GW), solar power (in steps up to 1.5 GW) and gas-fired capacity in the form of ICE (SPG, in steps up to 1.7 GW). The capacity of the renewables are added as additional capacity to the initial generation fleet of 11 GW, whereas the total amount of gas-fired capacity is kept constant (3.4 GW). The core purpose of the Dispatch Simulator is to raise awareness on the impact and value of adding SPG to the generation mix. The first reason why SPG capacity is replacing CCGT capacity is that it is realistic to assume that a new investment in thermal capacity is done in order to replace retiring capacity. Second, it is also reasonable to assume that an investment in new capacity should contribute to lowering the production costs. Therefore, in order to more objectively show the actual value of SPG and its flexibility, it replaces part of the CCGT capacity, on a one-to-one basis. This way it can be clearly demonstrated how SPG actually lowers the production costs to a greater extent than what CCGT is capable of. The key messages, as further described in chapter 4, as well as the benefits of SPG for power producers are in focus. The second objective of the Dispatch Simulator is to demonstrate how increasing amounts of intermittent, renewable power wind and solar affect the power system. On one hand, such capacity does not consume fuel, nor emit emissions, but on the other hand, new challenges arise. Foremost, the intermittent nature of such generation capacity creates an increased need for maintaining adequate power system reserves (as further described in chapter 3.2). As more and more intermittent capacity is added, the user will also realize the issues related to utilizing all the renewable energy produced. In order to so, the thermal power plants are forced to reduce their output considerably and even shut down units. Those power plants are then later forced to quickly increase their output again when the renewables at some point are producing less energy. This reflects the current situation in e.g. Germany well (How to 2013). The Dispatch Simulator features typical weather conditions, with varying wind speeds and solar radiation, as well as cloudy periods. The wind speed and solar radiation always have the same pattern in the Dispatch Simulator. The reason for this is that it enables the user to do direct comparisons between using CCGT and SPG for balancing. Even though randomised weather patterns would be more realistic, they would also make the assessment of the value of flexibility more complex for the user.

14 Status: Finalised Internal Document id: DBAD a Page 14 (25) As input for the wind and solar profiles used in the simulator, real historical data provided by Red Eléctrica de España (2014), the transmission agent and operator of the Spanish electricity system, is used. Spain is a very relevant choice, since its national power generation fleet has high shares of both wind and solar power (EIA 2011). Moreover, Red Eléctrica de España states that they are able to handle the variability in output from the renewables without importing any balancing energy (Klimstra & Hotakainen 2011: 105). Hence, the Spanish power system is an excellent source of input for the Dispatch Simulator, which does not feature any importing possibility at all. Based on the conditions in Spain on the 21 st of May 2014, a rather typical day in terms of weather conditions, the capacity factors of the Spanish wind and solar capacities, respectively, have been calculated. The capacity factor is the ratio between the actually produced electric energy and the theoretical, maximum amount of electric energy which could have been produced during that day. These real capacity factors are used in the simulator as basis for calculating the output from its wind and solar capacities. The load profile describes the consumption of electricity over time. In the Dispatch Simulator, the load profile is characteristic for most countries, with one peak in the morning and another one in the early evening. Finally, during the night hours, the demand is considerably reduced. (Klimstra & Hotakainen 2011: ) Also the load profile used in the simulator is based on real historical data from the Spanish nationwide grid operator Red Eléctrica de España (2014). The demand variations during the same day as earlier mentioned, the 21 st of May 2014, form the basis for the load curve in the simulator. Lastly, the Dispatch Simulator has a pricing system for CO2 emissions in place. This is based on the EU Emissions Trading System where there is a price tag for each tonne of emitted CO2 (EC 2013: 2). The price during has mostly been between five and six euro/tonne (EEX 2014) Power generation fleet performance Since the optimisation of the whole power generation fleet is the main objective for the user of the Dispatch Simulator, this chapter will provide a fairly detailed review of the key parameters of the different power plant technologies. The selection of power plant technology is based on a large number of different factors, such as capital investment costs, fuel costs, operation and maintenance costs, as well as subsidies, environmental permits and construction times (Engblom 2014: 23; Klimstra & Hotakainen 2011: ; Klimstra 2014: 16). However, since the Dispatch Simulator focuses exclusively on operative aspects, this chapter will only review the relevant parameters from a dispatching point of view.

15 Status: Finalised Internal Document id: DBAD a Page 15 (25) These operational parameters include: Unit size Fuel cost Fuel CO2 content Minimum load Start time (warm and cold) Stop time Ramp time at start (warm and cold) Ramp time while running Nominal efficiency Part-load efficiency Minimum up-time Minimum down-time Start-up maintenance impact Start-up fuel consumption First of all, the unit sizes of the different technologies vary considerably. This has a great impact on both the nominal efficiency and the part-load efficiency (as explained in chapter 4) (IEA 2014a: 172). In order to keep the generation fleet in the Dispatch Simulator perspicuous, the capacity of each thermal power plant technology is divided into blocks, which each represent several actual units (turbines or engines). As for all the remaining parameters mentioned above, the values used in the simulator are representative for typical, modern technology. They are based on a substantial review of literature, including but not limited to: Engblom (2014: 13), IEA (2014a: ; 2014b: ), Klimstra & Hotakainen (2011: ), Klimstra (2014: ), van den Bergh et al. (2013: 3), as well as data available in the Plexos software. In addition, several interviews with experts in Wärtsilä Power Plants have served as input. Still, it is worthwhile noting that the performance data remains an estimation, since the variations between different subtypes of technology is considerable. The parameters are presented in Table 1.

16 Status: Finalised Internal Document id: DBAD a Page 16 (25) Table 1 Generation fleet performance (Hultholm 2013b: 2). 5.3 Operation philosophy The baseload in the Dispatch Simulator is catered for by power plants with relatively low operational costs, i.e. nuclear, hydro and coal-fired assets which create the foundation for a stable system. Both nuclear and coal plants are normally large and utilise cheap fuel, but are on the other hand not suitable for compensating for sudden load changes or variations in the output from the renewables (Klimstra & Hotakainen 2011: ). Hydro capacity can be used for different purposes in a power system, depending on whether there is a reservoir available. With a reservoir, surplus energy can be used for pumping water to the storage and then later releasing it on demand. Such hydro capacity can be used for balancing purposes, while it otherwise is an intermittent power source. Natural conditions, such as seasonal droughts, also affect the availability of all kinds of hydro power. (Klimstra & Hotakainen 2011: ; Klimstra 2014: 110; Singh 2008: 18 91) In the Dispatch Simulator, the whole hydro capacity has been simplified to represent a combination of different kinds of hydroelectric plants: run-of-the-river with no storage, reservoir hydro and pumped hydro storage. The output of the hydro capacity in the simulator automatically follows the power demand and thus contributes to the system balancing. The maximum output is limited to 75% of the installed capacity, in order to reflect the fact that the availability of water varies (Klimstra 2014: 110).

17 Status: Finalised Internal Document id: DBAD a Page 17 (25) Finally, the gas-fired power plants with fast dynamics support the power system in the simulator by handling the remaining load fluctuations and balancing the renewables. Gas is a more expensive fuel than coal, but is utilised in more flexible plants which are therefore very suitable for balancing purposes (IEA 2014a: 263; Klimstra 2014: ). The above described operation philosophy of all the different power plant technologies is very much in line with e.g. the Spanish utilisation of its whole generation fleet (Klimstra & Hotakainen 2011: 105). So to sum up the operation philosophy described so far, the Dispatch Simulator basically utilises the merit order principle described in chapter 3.2. Hence, the generation assets are run in the following order: wind & solar, hydro, nuclear, coal, CCGT and SPG. The next chapter describes how this works in practice in the simulator. Chapter 6 Operation of the Dispatch Simulator After the user has selected potential additions to the generation fleet, as described in chapter 5.1, the simulation can begin. The user will function as a power system dispatcher, i.e. manage the whole system both in terms of stability and feasibility. 6.1 Demand vs. supply The Dispatch Simulator runs a 24-hours scenario, during which typical variations in power output from the renewable energy sources and load swings occur. As explained in chapter 3.1, power supply must continuously be matched with power demand in order to keep a power system stable. In order to simplify this balancing act, the Dispatch Simulator automatically matches the online capacity with the actual load. This occurs in accordance with the merit order principle described in chapter 5.3. Next, the simulator automatically distributes the load evenly between the blocks within each power plant technology. Hence, when considering the thermal power plants, the simulator first strives to operate the online coal blocks at highest possible output. After that, the output of the online CCGT blocks is set as high as possible. Finally, if the user has chosen to include SPG capacity, those blocks are the last to be assigned load. In other words, the SPG blocks are primarily used for continuously balancing the system and providing the system reserves. In practice, this means that the main task of the user is to ensure that a sensible amount of units are online all the time. When the online capacity starts be close to its full output, the user needs to start up more capacity to be able to match the demand. On the contrary, if the actual output of the total capacity online is low, the efficiency of the plants will also be low (as presented in chapter 4). Therefore, the user should shut down some of the online units in order to increase the output of the others and thereby increasing the efficiency. In case the power demand exceeds the total output of the capacity online and the user neglects to start up more capacity, load shedding will take place (as explained in chapter 3.1). In the simulator, the load shedding happens in two steps as the frequency decreases (49.8 Hz and 49.7 Hz), which is slight simplification of a typical setup (Blume 2007: ). Nevertheless, the frequency

18 Status: Finalised Internal Document id: DBAD a Page 18 (25) response in the Dispatch Simulator is modelled based on an actual frequency deviation study by the National Grid (2009). The simulator also calculates a penalty for the load shedding and adds it to the total generation costs. The costs imposed by loss of load varies considerably, e.g. based on the customer type and the amount of reduction, and can be anywhere in the range of $1000/MWh (Nguyen & Le 2013: 682) to $10 000/MWh (Philpott & Pritchard 2013: 7; Staheli 2005: 2 3). A fixed, reasonable amount of 1 MEUR / load shedding event (250 MW) is used in the simulator. If the user manages to increase the available capacity quickly enough, the dropped load will be restored by the Dispatch Simulator. However, ultimately, if no corrective actions are taken by the user and the situation is allowed to escalate even further, a full-scale blackout will eventually occur. This effectively ends the simulation. Oppositely, if the power demand is lower than the supply, the frequency will begin to increase. In this case, if the output of the online capacity cannot be further reduced, the Dispatch Simulator forces the user to react by shutting down some of the capacity. In order to simplify the simulator, severe over-frequency scenarios are this way avoided. 6.2 Decisions and control In accordance with the operation philosophy explained in chapter 5.3, the nuclear and the hydro capacity cannot be manually controlled in the Dispatch Simulator. Moreover, wind and solar power are so called non-dispatchable sources, since their output is at all times dependent on the prevailing weather conditions. That leaves the three thermal power plant technologies: coal-fired capacity, CCGT and SPG. First of all, the coal-fired blocks in the simulator are manually started and stopped. Since coal-fired power plants typically have very large units and thus a big impact on the whole power system, the coal blocks in the simulator can be operated in two different ways: spinning mode and efficiency mode. The former means that the allowed operation interval will be 40-80% of the nominal output of the block, i.e. a reserve capacity which can be manually activated, is maintained until further notice. The latter, efficiency mode, sets the available output range to % of the nominal output of the block. This setup allows the user to decide to which extent the coal-fired capacity participates in the provision of power system reserves. Second, also the CCGT blocks in the Dispatch Simulator are manually started and stopped. As CCGT power plants are typically used for balancing purposes and therefore operate over a relatively wide load range, the CCGT blocks in the simulator only have a single mode (with an output range of %) (DNV KEMA 2013: 15; Klimstra & Hotakainen 2011: 105; Redpoint 2013: 10). Third, the SPG blocks in the simulator are automatically controlled. When there is no other thermal capacity (coal and CCGT) left in the system to meet the demand, SPG blocks are automatically dispatched. The first reason why this is done automatically is the size of the blocks. As the SPG blocks are considerably smaller than the other ones, it would require a lot of repetitive activity from the user to manually control them. Second, the start-up time is so short that it would be virtually impossible for the user to fully benefit from it, considering the pace of the simulation (the 24 hour scenario elapsing in a couple of minutes).

19 Status: Finalised Internal Document id: DBAD a Page 19 (25) Correspondingly, when the demand decreases, or other capacity becomes available, the SPG blocks are the first to be unloaded. The blocks are then automatically unloaded one by one, returning to provision of non-spinning reserve. 6.3 Monitoring and information The Dispatch Simulator features a number of tools supporting the decision process of the user. The intention is that these will be educative, as well as contribute to making the user experience more interesting and interactive. Furthermore, the level of information richness of the tools varies, in order to cater for the needs and interests of different groups of users. First of all, there are pop-up messages which deliver information on two different levels: general information and critical notifications. The informative pop-ups are primarily intended for firsttime users and briefly review the current situation, as well as provide advice on how to proceed. In case SPG is included in the generation mix, these messages also highlight how SPG contributes to the optimisation of the system in different situations (i.e. the key messages defined in chapter 4). If the overall state of the power system becomes critical for some reason, a pop-up urging the user to take the necessary actions will emerge. Second, there is a message centre in the Dispatch Simulator. On one hand, all the received pop-up messages are stored here as a backlog to which the user can return at any point. In addition, feedback on the current power system reserve levels and plant efficiencies is provided. Third, the load curve for the 24-hour period is visible at all times, corresponding to a very accurate load forecast. The user can follow how the simulation progresses and how the load varies over the course of the simulated day. Fourth, there are a number of power system meters which are providing real-time information for the user. The power balance meter indicates how well demand and supply are matching, and should hence ideally be centred at its equilibrium position. Correspondingly, the frequency meter reflects the power system balance (as explained in chapter 3.1). A constant frequency, close to the nominal value of Hz implies a perfect stability. Additionally, there are three meters showing the current power demand, the amount of capacity available for upwards regulation (Reg-Up) and the amount of capacity available for downwards regulation (Reg-Down). Reg-Up describes how much the currently online capacity is able to further increase its current output, whereas Reg-Down expresses how much that same capacity is capable of reducing its output. Fifth, a number of real-time generation cost meters continuously supply the user with information on how well the dispatching is progressing from an economical point of view. The main one calculates the current generation cost for the whole generation fleet. In addition, the user can access the average generation cost so far in the simulation, as well as a cost breakdown of the different elements: continuous fuel consumption, start-up costs (fuel consumption and maintenance impact), CO2 allowances and possible load shedding penalties. Sixth, the dispatch centre gives more detailed information on how the simulation has gone so far. This report includes continuous monitoring of the power balance and frequency, the generation

20 Status: Finalised Internal Document id: DBAD a Page 20 (25) costs and the CO2 emissions, as well as the generation shares and load factors of the individual power plant technologies. Chapter 7 Outcome of the Dispatch Simulator Power systems need to be reliable, affordable and sustainable (IEA 2014a: ; IEA 2014b: 2 85; Klimstra 2014: ). Hence, the evaluation criteria of a simulation in the Dispatch Simulator are power quality, cost of produced electricity and caused CO2 emissions. In terms of power quality, the stability of the frequency is an excellent measure of the condition of a power system, i.e. its reliability (Ray 2007: 33). Ideally, the frequency should be constant and maintaining its stability is the main goal in a power system. (Klimstra & Hotakainen 2011: 78) In the Dispatch Simulator, the outcome of the power quality is evaluated based on a function that accumulates penalties on a stepwise basis when the frequency is off its nominal value Hz. The penalty gets increasingly severe the more the frequency deviates from its nominal value. Based on the accrued penalty, a reliability score is calculated. Ideally, it should be 100%, corresponding to minimal frequency deviation over the course of the whole simulated day. The affordability in the simulator is based on the average generation costs for the whole generation fleet during the simulation. The elements on which this costs is based are the continuous fuel consumption, the start-up costs (fuel consumption and maintenance impact), the CO2 allowances and possible load shedding penalties. Finally, the sustainability is measured by the Dispatch Simulator in terms of average amount of caused CO2 emissions per produced unit of electricity, taking all of the generation fleet into account. In addition to the numerical feedback and rating on scale ranging from poor to excellent, the user also receives a brief written assessment, as well as some hints on how to further improve the results. The results of the three latest simulations are automatically stored, which allows the user to compares the outcomes of different generation mixes. Hence, the impact of large amounts of intermittent renewables and SPG can be clearly shown. Chapter 8 The impact of Smart Power Generation As shown by the outcome of the simulation, Smart Power Generation, provided by the ICEs, allows the existing power generation fleet to be operated more efficiently and economically. When SPG fills up the fluctuating net load, the coal and CCGT plants can be run flat-out at a high efficiencies. Consequently, one can choose to focus on optimising the power system with focus on cleaner or on more inexpensive energy, depending on the proportions of the fuels used for the baseload. If the CCGT capacity, which has low CO2 emissions but high fuel costs, is used for providing a large